Prior-Guided Objective Detection and a 1993–2025 Monthly Record of the Northwest Atlantic Shelf-Slope Front
Shiliang Shan, Francois Rivest, Blair GreenanThe Northwest Atlantic Shelf-Slope Front (SSF) separates colder, fresher shelf water from warmer, saltier slope water and influences shelf-slope exchange, coastal ocean variability, and ecosystem conditions. Fisheries and Oceans Canada’s Atlantic Zone Monitoring Program (AZMP) maintained an analyst-derived monthly SSF position dataset beginning in 1973, but routine production ended after 2017. We developed a prior-guided objective sea surface temperature gradient peak detector to provide a reproducible ARMOR3D-based extension of this record. At integer longitudes from 75° W to 50° W, the detector identifies local maxima of the poleward cooling gradient and ranks them using peak prominence and distance from a seasonally adjusted shelf-break prior. Fivefold blocked cross-validation over 1993–2017 selected equal thermal prominence and geographic weighting. Across 6950 out-of-fold comparisons, bias was 0.095° latitude, root mean square error was 0.497°, median absolute error was 0.236°, and 75.7% of positions were within 0.50° of the AZMP-derived SSF. Seasonal peak timing differed by 0.13–0.35 months among the western, central, and eastern regions, and annual correlations were 0.89, 0.74, and 0.69, respectively. However, Peak SSF annual standard deviations were only 49–56% of the AZMP values, so the two records are not directly homogeneous in variability amplitude. The 2018–2025 portion is an ARMOR3D-based extension that has not been independently validated. It reveals regional contrasts rather than a persistent domain-wide displacement. The 1993–2025 Peak SSF dataset provides a reproducible, updateable benchmark for frontal monitoring, shelf-slope exchange studies, ecosystem analysis, and ocean model evaluation.